This report is focused around Lost and Found data using the intakes and outcomes data received for 2019 and 2020. Its goal is to reflect everything we could learn about L&F from the available data and highlight things that would be useful to show but some/all data required for them are missing.

Report Structure

  1. KPIs: data points that indicate how good the shelter is doing on on L&F. They have numeric goals associated with them.
  2. Supporting data: data points that aren’t a goal themselves but serve as a proxy for improving a goal. For example, the method of RTH is not a performance indicator, but it helps identifying how RTHs take place. The number of strays found per ZIP code is not a metric to improve, but it shows where most strays are coming from to guide resource allocation.
  3. Data status: the state of the data received from the shelter, highlighting missing pieces and potential areas for improvement.
  4. Extra metrics: some ideas for additional L&F metrics and the data points they require.

Scroll down or use the table of contents on the left to navigate throughout the document. Most sections contain multiple tabs showing different facets of a data type. Most plots are interactive, meaning they include tooltips and allow hiding and showing parts and zooming in and out. If something went wrong, look for the house icon in the top right corner of each figure to reset.

KPIs

Yearly RTH Rates by Species

This section provides an overview of the RTH rate per year divided by species. RTH Rate is calculated as the portion of returned animals that came in as strays out of stray animals. Animals younger than 6 weeks are excluded (this could be easily changed).

Overall RTH Rate

This table covers all strays and RTHs. The rate for dogs remained fairly similar between 2019 and 2020, while for cats, even though these are few returns, the number was the same (49 compared to 51) although intake was cut down in half.

Species Year Strays RTH_Count RTH_Rate
Bird 2019 12 1 0.08
Cat 2019 3806 51 0.01
Cat 2020 1684 49 0.03
Dog 2019 6860 1861 0.27
Dog 2020 5287 1460 0.28
Other 2019 41 1 0.02
Other 2020 42 3 0.07

Field RTH Rate

This one only counts animals who came in as strays from the field (using subtype ‘FIELD’ or ‘FIELD OWN’). These are then split by RTH method between field (using outcome_subtype ‘FIELD’), shelter (all other values), and unknowns (missing value).

Cats show no difference here, and while the in-field return rate for dogs increase, it might be because there are much less unknown (missing) subtypes. Nevertheless, it has the benefit of showing a more accurate rate of in-field returns (7% in 2020).

Species Year Field_Strays RTH_Subtype Field_RTH_Count RTH_Rate
Cat 2019 265 Field Return 1 0.00
Cat 2019 265 Shelter Return 4 0.02
Cat 2019 265 Unknown 5 0.02
Cat 2020 331 Field Return 4 0.01
Cat 2020 331 Shelter Return 11 0.03
Cat 2020 331 Unknown 2 0.01
Dog 2019 1463 Field Return 53 0.04
Dog 2019 1463 Shelter Return 135 0.09
Dog 2019 1463 Unknown 292 0.20
Dog 2020 1631 Field Return 112 0.07
Dog 2020 1631 Shelter Return 345 0.21
Dog 2020 1631 Unknown 54 0.03

Shelter RTH Rate by Species

Here lies the improvement in cat RTH as seen in the overall table, whereas dogs’ rates stay pretty much the same.

Species Year OTC_Strays Shelter_RTH RTH_Rate
Cat 2019 1715 41 0.02
Cat 2020 560 28 0.05
Dog 2019 4804 1257 0.26
Dog 2020 2867 773 0.27
Other 2019 14 1 0.07
Other 2020 23 1 0.04

RTH Over Time

These three time series show the RTH rate per month, to show whether there were times with particularly high or low rates as well as the overall trajectory.

Overall RTH

Again, we start with all RTH (both field and shelter), for cats and dogs. For dogs, December and January have noticeable spikes in both years. For cats, the rate is ever-slowly increasing, and the spike in April 2020 is due to a small intake amount.

Field RTH (Dogs)

This is the same figure, but only counting field strays, and showing only dogs due to cats’ low numbers. The different lines split the rate of return by Field RTH or in-shelter RTH, so in Dec 2019, for example, 29% of field stray intakes were returned in the shelter (outcome subtype was not ‘FIELD’) and an extra 7% were returned in the field. RTH dogs with missing subtypes were included in the Shelter category.

The in-shelter RTH numbers zigzag a lot while moderately decreasing over time, while Field returns show a slow increase peaking around May-July 2020, before slowly decreasing again later in 2020.

Shelter RTH

This figure only counts strays who did not come from the field. The trend for dogs is similar to the overall trend, and

Stray Intakes

This section shows the number of stray intakes over time, as well as the breakdown of strays by field/shelter intake.

Stray Intakes by Month

Noticeable parts include the spikes in cats (May 2019) followed by a rapid decline stabilizing around March 2020, and a sharp drop in stray intakes in April 2020 (again, covid) followed by a slow increase towards end of year.

Stray Intake Subtypes

Across both years, 24% of strays come the field (counting Field and Field Own) and the vast majority over the counter. 5% of values are missing.

Money Saved by RTH

This could be another useful metrics to reflect the benefits of RTH over other outcome types. It takes into account three components:

  1. The number of stray intakes with RTH outcome.
  2. The daily cost of care.
  3. The average difference in length of stay (in days) between strays with RTH outcomes and all other strays. This is shown in the table below – roughly 9 days for dogs and 6 for cats.

There were 1847 strays who got RTH in 2019 and 1451 in 2020. Assuming 50$ cost of daily care per dog, and given the length-of-stay differences, We can estimate that return-to-homes saved PACC \(1847*50*9=831,150\$\) in 2019 and \(1451*9*50=652,950\$\) in 2020.

Species Outcome Count Average_Length_Of_Stay
Cat Other Outcomes 2485 10.91
Cat RTO 100 5.00
Dog Other Outcomes 8025 12.37
Dog RTO 3298 3.37

Supporting Data

Stray Intake and RTH By Found ZIP

The following maps show stray intake and RTH rate by ZIP codes to highlight geographical patterns. The first and second tab are similar to previous metrics; the third tab, RTH Gap, shows the number of strays who were not returned home per ZIP code.

Stray Intake

The area around the shelter (85745) stands out even after the animals with a found location listing the shelter’s address were removed (about 150 of those). Generally, the numbers decrease the further the areas from the shelter.

RTH Rate

The one ZIP code further from Tuscon stands out – other than that, most areas are roughly similar.

RTH Gap

This combines the other two tabs to highlight where most additional RTH potential exists. Except for the shelter’s ZIP code, the areas adjacent to it all stand out as more strays are found in them with roughly similar RTH rates. In the areas to the Northwest, 743 and 653, the RTH rates (visible in the middle tab) are lower than other close areas, which suggests a greater room for improvement.

RTH Method

This could be a useful field to track, given that it reflects how an RTO was achieved. The table below shows the outcome subtype for RTH animals, where some values seem to reflect RTH method, excluding the values of walk-in and field which were used 90%+ of the time (hence the low numbers below). Fewer value options might help improving data collection on this field.

Year str_to_title(outcome_subtype) N
2019 Phone 58
2019 Web 40
2019 Facebook 18
2019 Chip 15
2019 Tag Id 10
2019 Otc 9
2019 Letter 7
2019 Next-Door 7
2019 Id Tag 5
2019 Craigslist 2
2019 Petharbor 1
2020 Phone 52
2020 Web 49
2020 Facebook 36
2020 Chip 34
2020 Otc 16
2020 Letter 14
2020 Neighbor 10
2020 Next-Door 7
2020 Other 7
2020 Email 5
2020 Petharbor 3
2020 Pawboost 2
2020 Tag Id 2
2020 Id Tag 1

Distances From Home

This section uses data received with the help of Josh @ Chameleon with the crossing and owner addresses of all RTO dogs from 2019 to date (Oct 2021).

Across this time period there were 4919 RTO outcomes. Out of those, 207 had the shelter address as crossing and 130 shelters had an unknown owner address, so they were removed. 15 more animals were removed for geocoding errors (finding their owner on the east coast or random states). 143 animals that had their owner address listed in Ajo, AZ were also removed, because it was ambiguous whether the cross streets were in Tuscon or Ajo. These can be explored separately if needed. 26 animals with the exact same crossing and shelter address were removed.

This left 4390 animals for whom distances between crossing and owner addresses were found and analyzed below. For each dog, the listed intake address and owner addresses were geocoded using Google’s geolocation service, and then the distance between the two points was calculated.

The distribution of distances traveled by these dogs is shown in the following figure. It appears that the majority of dogs remain close to home

To summarize this data in a concise way, we can look at the % of dogs who walked up to a block, a mile, five miles, or more. The following table shows the % of dogs falling within each distance category.

Overall, 8% dogs were found within a block away (~500 ft) from home and an additional 55% were found up to a mile away, making a total of 63% dogs found within a mile from home. 16% of dogs were found more than 5 miles away.

This table was derived separately for each of the years in the data, and the results are pretty similar (not shown).

Distance.Category Num_Animals Ratio
5+ miles 692 15.8%
1-5 miles 941 21.4%
More than a block, Less than a mile 2415 55%
Up to a block 342 7.8%

I also wanted to see if there was a difference between dogs coming in from the field versus over the counter. The table below shows the distances grouped by these two, where subtypes ‘FIELD’ and ‘FIELD OWN’ appear under Field and ‘OTC and OTC OWNED’ under OTC. The overall % under 1 mile is similar in both groups – 61% in the field and 64% OTC, but one difference is that 5% of OTC animals were found within a block as opposed to 12% for field intakes.

Distance.Category Field OTC
5+ miles 16.6% 15.2%
1-5 miles 22.5% 20.8%
More than a block, Less than a mile 48.5% 58.8%
Up to a block 12.3% 5.1%

The following table also shows the number of dogs and their average and median distances traveled when dividing them into field / OTC. The averages are higher in both cases because of the few outliers who went far away from home, so the median might be a better representation of the typical animal in this case, as it is less sensitive to these few outliers, and for all cases it is about 0.6 miles.

Subtype Count Average.Distance Median.Distance
Field 1498 2.48 0.62
OTC 2660 2.31 0.59
Other 232 2.51 0.56

Data Status

  1. Frequently missing values:

    • Intake reason is NA for 12191/40074 animals.
    • Outcome ZIP code: 23,644 NAs.
    • Found ZIP code: 5992 NAs.
    • Outcome subtype: 3682 NAs.
    • Outcome condition: 16063 NAs.
    • Intake subtype: 1816 NAs.
    • Weight: 1436 NAs.

  2. Intake subtype has 7 values assigned to less than 20 animals, could be removed to simplify.

  3. Outcome subtype - the presence of ways to indicate how a return was achieved stands out, but they are not used frequently. Here too there are multiple values that are rarely in use, which could be great for tracking how returns happen in the shelter.

Extra Metrics

Other things we could show if we had the data for it:

  1. Prevalence of microchips across town (for example, are there areas from which more animals come in without chips?) and the RTH rates for animals found with/without chips.
  2. Reclaim fees (could be a yes/no to track fee waiving).
  3. Number of public found reports and successful RTH by public (if this data is accessible to the shelter; these two would allow showing $ saved by public RTH).